Browsing: Guides & Tutorials

TL;DR The enterprise AI market makes more sense as an architecture map than as a vendor ranking. Microsoft, Google, Salesforce, OpenAI, Anthropic, Meta, Mistral, NVIDIA, AMD, Intel, Dell, HPE, Cisco, Lenovo, Supermicro, Arista, and Broadcom are not competing inside isolated categories. They are competing for control points that span user workflows, model distribution, inference runtimes, private infrastructure, network fabrics, and accelerator economics. The most important question is not which company is the biggest shark. It is which company controls the boundary your organization cannot easily replace. A SaaS platform may control business context. A model provider may control intelligence quality…

Agent incidents do not always look like outages. A traditional service incident might show up as latency, errors, saturation, failed deployments, or unavailable endpoints. Those still matter for AI agents, but they are not the whole story. An agent can be technically available and still create an operational incident. It might expose restricted data, call the wrong tool, rely on stale context, bypass an approval path, retry an expensive workflow, modify the wrong object, or follow malicious instructions embedded in a document. That is why rollback for AI agents needs more than a redeploy button. For production agents, rollback means…

Introduction Enterprise inference standardization often begins with a benchmark chart. That is usually where the first mistake occurs. A team runs one model on one GPU configuration, increases concurrency until the system saturates, records the highest tokens-per-second result, and declares a winner. The result may be technically repeatable while still being operationally irrelevant. It says little about how the runtime behaves under production latency objectives, mixed prompt lengths, model updates, node failures, autoscaling events, cold starts, or security controls. The comparison is also frequently framed incorrectly. NVIDIA NIM, NVIDIA Triton Inference Server, vLLM, and TensorRT-LLM are not four equivalent inference…

TL;DR NVIDIA Run:ai can turn a shared Kubernetes GPU cluster into a governed multi-tenant platform by organizing workloads into departments and projects, assigning guaranteed GPU quotas per node pool, and allowing controlled over-quota use when capacity would otherwise remain idle. The design depends on four controls working together: Quota establishes the resource entitlement for a department or project. Fairshare determines how unused capacity is distributed. Priority orders workloads within the relevant scheduling queue. Preemptibility determines whether borrowed capacity can be reclaimed. Production workloads should normally run from quota-backed projects as non-preemptible workloads. Development and experimental workloads can use preemptible over-quota…

TL;DR Choose a deterministic workflow when the path is known, repeatability matters, latency must stay low, and every transition needs to be explainable. Choose one agent when the task is variable or open-ended, but the work still fits within one domain, one security boundary, and one accountable product team. Choose multiple agents only when specialization, organizational boundaries, security isolation, parallel work, or context separation create measurable value that a single agent cannot deliver reliably. The practical default is not “agentic.” It is the lowest-complexity architecture that meets the requirement. In many enterprise systems, the strongest design is hybrid: a deterministic…

Introduction Platform transformation is usually planned as a sequence of discoveries, upgrades, replication jobs, test migrations, cutovers, and decommissioning activities. Recovery is often treated as a separate operational concern that will somehow continue working while those activities occur. That assumption is dangerous. During a VMware upgrade, hypervisor migration, or multiyear platform replacement, the environment contains more versions, more trust relationships, more temporary tooling, more replication paths, and more partially migrated applications than it does during steady-state operations. The transformation program is deliberately changing the same storage, networking, identity, management, and compute dependencies that the existing recovery plan assumes will remain…

Private AI does not always fail because the model is wrong. It often fails because the organization never defines the control point. The first team deploys a model. Another team connects an agent. A developer experiments with a hosted LLM. A business unit wants internal document search. Security asks which data sources the agent can reach. Finance asks who owns the token usage. Operations asks where the logs live. Platform engineering asks why every project is building a different path to inference. That is the enterprise private AI problem Nutanix Enterprise AI is trying to solve. Nutanix Enterprise AI is…

TL;DR NVIDIA GPU monitoring needs more than a utilization chart. A production design should collect device telemetry with DCGM Exporter, scrape it with Prometheus, visualize fleet and workload behavior in Grafana, and alert on conditions that require action. The runbook must also preserve per-pod context, control metric cardinality, distinguish low utilization from genuine performance problems, and capture enough evidence to support escalation. The safest implementation path is to use the DCGM Exporter already deployed by NVIDIA GPU Operator when one exists. Deploy a standalone exporter only when the Operator is not managing GPU telemetry. Validate the exporter, ServiceMonitor, Prometheus target,…

TL;DR Deploying NVIDIA vGPU on VMware vSphere is not simply a matter of installing a driver and attaching a virtual PCI device. The server, physical GPU, ESXi build, NVIDIA vGPU release, guest operating system, vGPU profile, and licensing model must all form a supported combination. A reliable deployment sequence is: Validate the complete hardware and software compatibility matrix. Configure the required BIOS, IOMMU, SR-IOV, and display settings. Install the NVIDIA vGPU Manager components on the ESXi host. Configure the host graphics mode as Shared Direct. Add an appropriate vGPU profile to a powered-off VM. Install the matching NVIDIA vGPU guest…

TL;DR NVIDIA NeMo Guardrails should sit in the controlled request path between the application and the production LLM endpoint, not beside it as an optional validation service. The application sends requests through an authenticated gateway to the guardrails service. Input rails inspect the request, retrieval rails inspect RAG context, execution rails constrain tool use, the main request is sent to NVIDIA NIM, and output rails validate the model response before it reaches the caller. A production implementation also needs controls that do not appear in a basic demo: no direct application path to NIM, explicit fail-open and fail-closed decisions, separate…